[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com> Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
co-authored by
Claude Opus 5
parent
1085b64425
commit
651a88c09a
@@ -22,8 +22,7 @@ def test_sanitize_message():
|
||||
|
||||
class TestSanitizeMessageFilePaths:
|
||||
"""sanitize_message should also strip file paths and traceback
|
||||
frames, not just memory addresses - see #31683.
|
||||
"""
|
||||
frames, not just memory addresses - see #31683."""
|
||||
|
||||
def test_strips_traceback_style_frame(self):
|
||||
msg = (
|
||||
@@ -45,8 +44,7 @@ class TestSanitizeMessageFilePaths:
|
||||
|
||||
def test_strips_single_parent_container_path(self):
|
||||
"""Regression: /app/server.py and /workspace/server.py (common in
|
||||
container deployments) were missed by the original {2,} quantifier.
|
||||
"""
|
||||
container deployments) were missed by the original {2,} quantifier."""
|
||||
assert "/app/" not in sanitize_message("Error in /app/server.py")
|
||||
assert "/workspace/" not in sanitize_message("Error in /workspace/server.py")
|
||||
|
||||
|
||||
@@ -24,8 +24,7 @@ from vllm.entrypoints.serve.exception_handling.handlers.validation import (
|
||||
|
||||
def _fake_request(log_error_stack: bool = False) -> SimpleNamespace:
|
||||
"""Minimal stand-in for a FastAPI Request - just enough for the
|
||||
handler to read req.app.state.args.log_error_stack.
|
||||
"""
|
||||
handler to read req.app.state.args.log_error_stack."""
|
||||
return SimpleNamespace(
|
||||
app=SimpleNamespace(
|
||||
state=SimpleNamespace(args=SimpleNamespace(log_error_stack=log_error_stack))
|
||||
@@ -36,8 +35,7 @@ def _fake_request(log_error_stack: bool = False) -> SimpleNamespace:
|
||||
|
||||
class TestValidationErrorParamFallback:
|
||||
"""Ensure `param` falls back to the Pydantic error's `loc` when no
|
||||
custom VLLMValidationError context is present.
|
||||
"""
|
||||
custom VLLMValidationError context is present."""
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
("error_type", "msg"),
|
||||
@@ -61,8 +59,7 @@ class TestValidationErrorParamFallback:
|
||||
async def test_param_fallback_does_not_crash_on_non_dict_error(self):
|
||||
"""Schemathesis fuzzing found that errors[0] isn't always a dict.
|
||||
The fallback must not crash in that case - it should just leave
|
||||
param as None instead of raising.
|
||||
"""
|
||||
param as None instead of raising."""
|
||||
exc = RequestValidationError(["some unexpected non-dict error"])
|
||||
|
||||
response = await validation_exception_handler(_fake_request(), exc)
|
||||
@@ -100,8 +97,7 @@ class TestCleanLocForParam:
|
||||
|
||||
def test_all_internal_falls_back_to_raw_join(self):
|
||||
"""If every loc segment looks internal, fall back to the raw
|
||||
dot-join rather than returning an empty string.
|
||||
"""
|
||||
dot-join rather than returning an empty string."""
|
||||
loc = ("function-wrap[__log_extra_fields__()]",)
|
||||
assert clean_loc_for_param(loc) == "function-wrap[__log_extra_fields__()]"
|
||||
|
||||
@@ -109,8 +105,7 @@ class TestCleanLocForParam:
|
||||
class TestValidationErrorDoesNotLeakServerPaths:
|
||||
"""FastAPI sets endpoint_file/line/function/path on the exception
|
||||
during routing, not in the constructor - so we set them manually
|
||||
here to reproduce the real leak. See #31683.
|
||||
"""
|
||||
here to reproduce the real leak. See #31683."""
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_handler_strips_endpoint_file_context(self):
|
||||
|
||||
@@ -245,8 +245,7 @@ async def test_multiple_lora_adapters(
|
||||
client: openai.AsyncOpenAI, tmp_path, qwen3_lora_files
|
||||
):
|
||||
"""Validate that many loras can be dynamically registered and inferenced
|
||||
with concurrently
|
||||
"""
|
||||
with concurrently"""
|
||||
|
||||
# This test file configures the server with --max-cpu-loras=2 and this test
|
||||
# will concurrently load 10 adapters, so it should flex the LRU cache
|
||||
|
||||
@@ -72,8 +72,7 @@ class MockModelConfig:
|
||||
@pytest.fixture(params=get_args(SupportedTask))
|
||||
def task_routes(request, monkeypatch) -> tuple[str, list[tuple[str, list[str]]]]:
|
||||
"""For each supported task, build an app with only that task's routers,
|
||||
extract all routes, and return the task name and routes.
|
||||
"""
|
||||
extract all routes, and return the task name and routes."""
|
||||
task = request.param
|
||||
# Enable development mode to register all routes (including dev-only routes).
|
||||
monkeypatch.setenv("VLLM_SERVER_DEV_MODE", "1")
|
||||
@@ -99,8 +98,7 @@ def task_routes(request, monkeypatch) -> tuple[str, list[tuple[str, list[str]]]]
|
||||
|
||||
def test_auto_discovered_protected_routes_require_auth(task_routes):
|
||||
"""For every auto-discovered route that starts with a guarded prefix,
|
||||
verify that authentication is enforced.
|
||||
"""
|
||||
verify that authentication is enforced."""
|
||||
task, routes = task_routes
|
||||
app = _create_app_with_mock_routes(routes)
|
||||
client = TestClient(app)
|
||||
@@ -134,8 +132,7 @@ def test_auto_discovered_protected_routes_require_auth(task_routes):
|
||||
|
||||
def test_auto_discovered_unprotected_routes_no_auth(task_routes):
|
||||
"""For every auto-discovered route that does NOT start with a guarded
|
||||
prefix, verify that no authentication is required.
|
||||
"""
|
||||
prefix, verify that no authentication is required."""
|
||||
task, routes = task_routes
|
||||
app = _create_app_with_mock_routes(routes)
|
||||
client = TestClient(app)
|
||||
|
||||
@@ -114,8 +114,7 @@ class TestHandlerOverrideIntegration:
|
||||
|
||||
def test_customer_script_functions_auto_loaded(self, monkeypatch, tmp_path):
|
||||
"""Test customer scenario: script functions automatically override
|
||||
framework defaults.
|
||||
"""
|
||||
framework defaults."""
|
||||
try:
|
||||
from model_hosting_container_standards.sagemaker.config import (
|
||||
SageMakerEnvVars,
|
||||
@@ -176,8 +175,7 @@ async def custom_sagemaker_invocation_handler(request: Request):
|
||||
|
||||
def test_customer_decorator_usage(self, monkeypatch, tmp_path):
|
||||
"""Test customer scenario: using @custom_ping_handler and
|
||||
@custom_invocation_handler decorators.
|
||||
"""
|
||||
@custom_invocation_handler decorators."""
|
||||
try:
|
||||
from model_hosting_container_standards.sagemaker.config import (
|
||||
SageMakerEnvVars,
|
||||
@@ -233,8 +231,7 @@ async def my_invoke(request: Request):
|
||||
|
||||
def test_handler_priority_order(self, monkeypatch, tmp_path):
|
||||
"""Test priority: @custom_ping_handler/@custom_invocation_handler
|
||||
decorators vs script functions.
|
||||
"""
|
||||
decorators vs script functions."""
|
||||
try:
|
||||
from model_hosting_container_standards.sagemaker.config import (
|
||||
SageMakerEnvVars,
|
||||
@@ -307,8 +304,7 @@ async def custom_sagemaker_invocation_handler(request: Request):
|
||||
|
||||
def test_environment_variable_script_loading(self, monkeypatch, tmp_path):
|
||||
"""Test that environment variables correctly specify script location
|
||||
and loading.
|
||||
"""
|
||||
and loading."""
|
||||
try:
|
||||
from model_hosting_container_standards.sagemaker.config import (
|
||||
SageMakerEnvVars,
|
||||
@@ -420,8 +416,7 @@ async def custom_sagemaker_invocation_handler(request: Request):
|
||||
|
||||
def test_handler_env_var_override(self, monkeypatch, tmp_path):
|
||||
"""Test CUSTOM_FASTAPI_PING_HANDLER and CUSTOM_FASTAPI_INVOCATION_HANDLER
|
||||
environment variable overrides.
|
||||
"""
|
||||
environment variable overrides."""
|
||||
try:
|
||||
from model_hosting_container_standards.common.fastapi.config import (
|
||||
FastAPIEnvVars,
|
||||
@@ -515,8 +510,7 @@ async def custom_sagemaker_invocation_handler(request: Request):
|
||||
|
||||
def test_env_var_priority_over_decorator_and_script(self, monkeypatch, tmp_path):
|
||||
"""Test that environment variables have highest priority over decorators
|
||||
and script functions for both ping and invocation handlers.
|
||||
"""
|
||||
and script functions for both ping and invocation handlers."""
|
||||
try:
|
||||
from model_hosting_container_standards.common.fastapi.config import (
|
||||
FastAPIEnvVars,
|
||||
|
||||
@@ -51,8 +51,7 @@ async def test_tokenize_then_chat_completion_with_image(
|
||||
local_asset_server,
|
||||
):
|
||||
"""Tokenize a multimodal message, then send the same message to chat
|
||||
completions. The chat completion must succeed (not 500).
|
||||
"""
|
||||
completions. The chat completion must succeed (not 500)."""
|
||||
image_url = local_asset_server.url_for("stop_sign.jpg")
|
||||
messages = [
|
||||
{
|
||||
|
||||
@@ -36,13 +36,11 @@ def test_should_include_usage_force_enables_continuous_usage(stream_options, exp
|
||||
class TestGetMaxTokens:
|
||||
"""Tests for get_max_tokens() to ensure generation_config's max_tokens
|
||||
acts as a default when from model author, and as a ceiling when
|
||||
explicitly set by the user.
|
||||
"""
|
||||
explicitly set by the user."""
|
||||
|
||||
def test_default_sampling_params_used_when_no_request_max_tokens(self):
|
||||
"""When user doesn't specify max_tokens, generation_config default
|
||||
should apply.
|
||||
"""
|
||||
should apply."""
|
||||
result = get_max_tokens(
|
||||
max_model_len=24000,
|
||||
max_tokens=None,
|
||||
@@ -53,8 +51,7 @@ class TestGetMaxTokens:
|
||||
|
||||
def test_request_max_tokens_not_capped_by_default_sampling_params(self):
|
||||
"""When user specifies max_tokens in request, model author's
|
||||
generation_config max_tokens must NOT cap it (fixes #34005).
|
||||
"""
|
||||
generation_config max_tokens must NOT cap it (fixes #34005)."""
|
||||
result = get_max_tokens(
|
||||
max_model_len=24000,
|
||||
max_tokens=5000,
|
||||
@@ -97,8 +94,7 @@ class TestGetMaxTokens:
|
||||
|
||||
def test_request_max_tokens_smaller_than_default(self):
|
||||
"""When user explicitly requests fewer tokens than gen_config default,
|
||||
that should be respected.
|
||||
"""
|
||||
that should be respected."""
|
||||
result = get_max_tokens(
|
||||
max_model_len=24000,
|
||||
max_tokens=512,
|
||||
|
||||
@@ -239,8 +239,7 @@ def test_request_logger_log_outputs_integration():
|
||||
|
||||
def test_streaming_complete_logs_full_text_content():
|
||||
"""Test that streaming complete logging includes
|
||||
full accumulated text, not just token count.
|
||||
"""
|
||||
full accumulated text, not just token count."""
|
||||
mock_logger = MagicMock()
|
||||
|
||||
with patch("vllm.entrypoints.serve.utils.request_logger.logger", mock_logger):
|
||||
|
||||
@@ -17,8 +17,7 @@ from vllm.entrypoints.serve.utils.sse_keep_alive import (
|
||||
def test_disabled_returns_same_generator():
|
||||
"""For 0, negative, and NaN intervals the exact same generator object must
|
||||
be returned (identity), proving no wrapper or per-chunk overhead when
|
||||
disabled.
|
||||
"""
|
||||
disabled."""
|
||||
|
||||
async def gen():
|
||||
yield "x"
|
||||
@@ -34,8 +33,7 @@ def test_disabled_returns_same_generator():
|
||||
@pytest.mark.asyncio
|
||||
async def test_keep_alive_during_silent_start():
|
||||
"""A silent start (queue wait / prefill) emits keep-alive comments before
|
||||
the first real chunk.
|
||||
"""
|
||||
the first real chunk."""
|
||||
release = asyncio.Event()
|
||||
|
||||
async def gen():
|
||||
@@ -57,8 +55,7 @@ async def test_keep_alive_during_silent_start():
|
||||
@pytest.mark.asyncio
|
||||
async def test_no_keep_alive_when_streaming_is_fast():
|
||||
"""If chunks arrive faster than the interval, no comment is emitted and all
|
||||
data is forwarded.
|
||||
"""
|
||||
data is forwarded."""
|
||||
|
||||
async def gen():
|
||||
for chunk in ["a", "b", "c"]:
|
||||
@@ -72,8 +69,7 @@ async def test_no_keep_alive_when_streaming_is_fast():
|
||||
@pytest.mark.asyncio
|
||||
async def test_keep_alive_between_chunks_does_not_drop_data():
|
||||
"""Keep-alives interleaved with real chunks must never drop or reorder
|
||||
data.
|
||||
"""
|
||||
data."""
|
||||
release_b = asyncio.Event()
|
||||
release_c = asyncio.Event()
|
||||
|
||||
@@ -139,8 +135,7 @@ async def test_upstream_cancelled_error_propagates():
|
||||
@pytest.mark.asyncio
|
||||
async def test_aclose_closes_upstream():
|
||||
"""Closing a wrapper that is mid-stream must complete promptly and close
|
||||
the upstream generator (regression for the earlier leaked-task hang).
|
||||
"""
|
||||
the upstream generator (regression for the earlier leaked-task hang)."""
|
||||
finalized = asyncio.Event()
|
||||
|
||||
async def gen():
|
||||
@@ -159,8 +154,7 @@ async def test_aclose_closes_upstream():
|
||||
@pytest.mark.asyncio
|
||||
async def test_client_disconnect_cancels_upstream():
|
||||
"""Cancelling the consumer mid-stream must cancel the pending anext and
|
||||
close the upstream generator so its cleanup runs.
|
||||
"""
|
||||
close the upstream generator so its cleanup runs."""
|
||||
finalized = asyncio.Event()
|
||||
|
||||
async def gen():
|
||||
@@ -187,8 +181,7 @@ async def test_client_disconnect_cancels_upstream():
|
||||
@pytest.mark.asyncio
|
||||
async def test_streaming_response_emits_comment():
|
||||
"""A keep-alive comment must reach the client through the real
|
||||
StreamingResponse used by the routers.
|
||||
"""
|
||||
StreamingResponse used by the routers."""
|
||||
release = asyncio.Event()
|
||||
|
||||
async def gen():
|
||||
@@ -227,8 +220,7 @@ async def test_streaming_response_emits_comment():
|
||||
async def test_streaming_response_disconnect_closes_upstream():
|
||||
"""On the ASGI 2.3 disconnect path used by the pinned Uvicorn, Starlette
|
||||
cancels streaming and the wrapper's finally must close the upstream
|
||||
generator.
|
||||
"""
|
||||
generator."""
|
||||
finalized = asyncio.Event()
|
||||
|
||||
async def gen():
|
||||
|
||||
Reference in New Issue
Block a user